332 Dr. T. Scheerer and Mr. W. Francis on some 



rnerated crystals, terminated at the extremities by an oblique 

 plane. Its dull surface, arising from a film of oxide, to- 

 gether with the fascicular arrangement, does not allow of a 

 more accurate crystallographical determination ; it is pro- 

 bable that they belong to the mono-dimetric system. Some 

 of the crystals, or rather of the bundles of crystals, were 

 above two inches long, and about two lines in thickness. The 

 analysis was performed in the mode usually employed for the 

 arsenic metals. Cobalt and iron were separated from each 

 other by approximate neutralization of the sulphate solutions 

 of the two metals, by dilution and then boiling them. 

 The result of the analysis was 



Sulphur 0'16 



Arsenic 36-02 



Cobalt 53-71 



Iron 10-05 



Copper 0-86 100-80. 



The peroxide of iron here obtained was found before the 

 blow- pipe to be perfectly free from oxide of cobalt; the oxide 

 of cobalt, however, contained a slight trace of iron, which 

 was separated by dissolving the oxide in hydrochloric acid, 

 and precipitating with an excess of ammonia. It may have 

 amounted, at the utmost, to a few milligrammes. 



To be certain that these crystals contained no admixture 

 of foreign substances, but were a pure chemical combina- 

 tion, the following experiments were made with other por- 

 tions of them. In the first place, the amount of arsenic, the 

 accurate determination of which was most essential, was 

 ascertained in two other additional cases, and found to be as 

 follows : 



1st, 36-4-1 per cent. 



2nd, 35-34 



approximating, therefore, to the quantity first found, 36'02. 



The sum of the oxide of cobalt and peroxide of iron, which 

 in the first analysis had amounted to 2-061 grammes, was, 

 in a second examination, for which a quantity differing from 

 the first only by 0*005 was employed, equal to 2-067 grammes. 

 In both experiments the oxides were kept at a red heat in 

 an open platinum crucible, until they no longer increased in 

 weight*. Lastly, the quantity of cobalt was again deter- 

 mined by a second trial, and found to be 54-90 ; whereas 

 the first gave it 53-71. The differences occurring between 

 these several results are certainly not such as to give rise to 



*. When a larger quantity of the peroxide of iron is heated with a smaller 

 quantity of the oxide of cobalt for a short time in a closed platina crucible, 

 the latter scarcely becomes more highly oxidized, and in analyses not 

 requiring extreme accuracy it may be admitted that no superoxide it 

 formed. 



